Associations between Forest Fire and Mexican Spotted Owls Jeffrey S. Jenness, Paul Beier, a n d Joseph L. Ganey ABSTRACT. In 1993, the US Fish and Wildlife Service listed the Mexican spotted owl (Strix occidentalis lucida) as threatened, in part because of the rising threat to its habitat from stand-replacing wildfires. In 1997, we surveyed 33 owl sites that, in the previous four years, had burned at various levels ranging from light controlled burns to stand-replacing fires. We compared owl occupancy and reproduction in these burned sites to 31 unburned owl sites with similar habitat and topography. Although unburned sites showed higher proportions of both occupancy and reproduction, the negative relationship observed between recent fire occurrence and owl occupancy rank was statistically weak (Test for Marginal Homogeneity, P = 0.110). Owls tended not to be present where pure pine stands (Pinus spp.) comprised a large proportion (38-85%) of burned sites, but no other factors relating to habitat or fire severity had a significant, biologically interpretable influence on occupancy rank. We suspect that relatively low-intensity ground fires, including most prescribed fires, probably have little or no short-term impact on Mexican spotted owl presence or reproduction, but we have no data on long-term effects of fire. We recommend proactive fuels-management treatments in areas not currently occupied by owls as a means of reducing fire risk in areas occupied by owls. Within areas occupied by owls, judicious treatments may be appropriate after caseby-case evaluations of potential benefits and risks within those sites. FOR. SCI. 50(6): 765-772. Key Words: Prescribed burn, occupancy, reproduction, stand-replacement, standmaintenance.
ILDFIRE HISTORICALLY H A S BEEN 3. major dis-
turbancc agent important in structuring southwestern forests (Swetnam 1990, Covington and moo re 1994a). Fire suppression, timber harvesting, and grazing have altered southwestern forests greatly over the past century, however (Marshall 1963, Covington and Moore 1994a, 1994b, Kolb et al. 1994). These activities have lnarkedly increased litter and combustible debris on
the ground, and denuty of understory \hrubs and small trees (Leopold 1924, Madariy and West 1983, Savage and Swetnarn 1990). I n turn, these conditions increase the potential for larger and hotter fires (Moody et al. 1992, Covington et al. 1994). Thus. management actions have altered the fire regime i n some southwestern forest types, such as t ~ ) mixed-conifer forponderosa pine (Pinur p o t ~ c l ~ r o sand a t \ , from one characterized by frequent, low-intensity
Jeffrey S. Jenness, USDA Forest Service, Rocky Mountain Research Station, 2500 S. Pine Knoll Drive, Flagstaff, AZ 86001 -Phone: (928) 526-4139; Fax: (928) 556-2130;
[email protected]. Paul Beier, School of Forestry and MerriamPowell Center for Environmental Research, Northern Arizona University, Flagstaff, AZ 8601 1 -Phone: (928) 523-9341;
[email protected]. Joseph L. Ganey, USDA Forest Service, Rocky Mountain Research Station, 2500 S. Pine Knoll Drive, Flagstaff, AZ 86001 -Phone: (928) 556-2156;
[email protected]. Acknowledgments: Funding for this project was provided by the USDA Forest Service Rocky Mountain Research Station through a cooperative agreement (28-C6-915) with Northern Arizona University. We thank T. Edwards, E. Ellicott, B. Gill, L. Koch, R. Koch, N. Nahstoll, and T. Story for collecting field data. For data and technical assistance, we thank B. Block, D. Bieber, P. Boucher, M. Cederholm, M. DeLaune, R. Duncan, G. Froehlich, T. Grubb, I. Harris, G. Helbing, R. King, C. May, M. McCluhan, S. Morley, R. Pope, T. Randall-Parker, D. Salas, G. Sheppard, T. Skinner, C. van Riper Ill, and L. Engelman. For reviews and comments, we thank P. Stacey, G. Amacher, E. Green and four anonymous reviewers. Manuscript received September 9, 2002, accepted July 23, 2004.
Copyright 0 2004 by the Society of American Foresters For.c,.\/Scictitr 50(6) 20(M
765
fires, to one characterized more by infrequent, highintensity fires. The Mexican spotted owl (Strix nccidetitalis Iucidc~)inhabits mixed-conifer and pine-oak (Qiret.c~rsspp.) forests antl rocky canyonlands in the southwestern United Slates and Mexico (GutiCrrez, et al. 1995, USDI Fish antl Wildlire Service 1995). This owl thus hi^^ a long evolutionary history in forests structured by frequent, low-intensity fires. The Mexican spotted owl was listed as a threatened species in 1993, based on historical antl ongoing habitat alteration resulting from timber-management practices and the threat of additional habitat loss from stand-replacing wildfires (USDI Fish and Wildlife Service 1993). Large-scale standreplacing fires occurring throughout the owl's range could harm the owl by reducing or eliminating roosting, nesting, and foraging habitat (Sheppard antl Farnsworth 1997). The indirect effects of high-intensity wildfire on Mexican spotted owls are not known, unfortunately. and descriptive data are lacking on fire behavior at sites used by owls and how owls respond to fire-induced changes to their habitat (but see Sheppard and Farnsworth 1997). Fire may affect spotted owls either directly or inclirectly. Mexican spotted owls typically nest and roost in structurally-complex, diverse forests with a variety of age and size classes, a component of large trees, often with many snags antl down logs and relatively high basal areas and canopy closures (Ganey and Dick 1995, Ganey et al. 1999). Fire directly affects these habitat components either by destroying them, as is typical in a stand-replacing fire, or enhancing them by creating snags, thinning densely packed stands, and reducing competition for water and nutrients. Fire may indirectly affect these habitat components by reducing l'uel loads and thereby reducing the chance for stand-replacing fires. Spotted owls may select habitats partially based on prey availability (Verner et al. 1992, Ward and Block 1995), so fire can affect owls indirectly by altering habitat conditions for their prey. Ward and Block (1995) summarized data suggesting that reproductive success of Mexican spotted owls was influenced by total prey biomass rather than the relative abundance of any particular species. Some Mexican spotted-owl prey species show a decline or mixed response after fire, but many species. especially deer mice (Perortrysc~istizc~r~ic~ul~~tirs), increase in abundance after fire (Tevis 1956. Buech et al. 1977, Wirtz et al. 1988). Early successional specialists (such as the deer mouse) and species that require open habitats with well-developed herbaceous understories (such as pocket gophers [Thomotny.~ spp.] or microtine voles [Mio-otus spp.]) benefit h-om intense stand-replacing fires, whereas species that require dense canopies or ground cover decline (McGee 1982, Ward and Block 1995). Seed-eating species would find a sudden increase in their food supply when annual grasses come in. The effects of fire on prey biomass are varied, but generally appear to be short-lived (Bock and Bock 1983. Martell 1984, Kirkland et al. 1996). Deer mouse populations appear to universally increase after fire (Wirtz 1982,
Kaufman et al. 1988, Schwilk and Keeley 1998). Total prey biomass after fire appears to increase in some areas and decrease in others. Thus, overall effects of fire on prey biomass also are equivocal. The reduction in ground cover caused by fire could benefit the owl by leaving the prey more exposed, however. thus increasing prey availabilily. To assess potential impacts of fire on Mexican spotted owls. we compared the presence and reproductive success of owls between recently burned sites and similar unburned sites (we use "sites" to rel'er to areas with a histo~yof recent occupancy by owls, loosely analogous to owl lerritories). We also collected data on severity and extent of burn, topographic characteristics, and dominant foresl cover types in burned sites to compare the importance of fire to other possible predictor variables. Our research questions were: (1) Does fire in an owl site influence spotted owl occupancy and reproduction 1-4 years after the fire. compared to unburned sites: and (2) Given a fire within an owl site, does severity and extent of fire, dominant cover type, or topography within the burned site influence spotted owl occupancy and reproduction'? We also briefly describe fire extent (percent of forest cover types burned) for recenl l'ires in owl sites.
Methods Study Area and Site Selection Using information gathered from USDA Forest Service (USFS) offices, we selected owl sites in Arizona and New Mexico that burned during 1993-1996. These included sites burned by prescribed and prescribetl-natural fire (naturally occurring fires thal burn in preplanned locations, which are allowed to burn while they meet management objectives), and wildfire. For each burned site, we selected a paired unburned site ( i t . , no evidence of recent burn) within 12 km that had similar topography and l'orest cover types. We selected paired sites based on recornmendations from local USFS and other research biologists, drawing on their experience and knowledge of the area. We initially selected 32 pairs of burned and unburned sites. Ground surveys revealed no evidence of recent fire in one of the "burned" sites, however, and extensive burned areas in two "unburned" sites. Thus, our sample consisted of 33 burned sites and 31 unburned sites, including 29 pairs of burned/ unburned sites, distributed among nine geographic areas and four national forests (Fig. 1). Prior knowledge of occupancy patterns on these sites varied greatly in quality and quantity, but all sites had a record of, at minimum, either a pair of owls in a single year or a single owl in at least two years, between 1990 and 1996. USFS survey histories were available for 53 of the 64 sites. Historic occupancy rates for these sites are calculated as the number of years of occupancy recorded divided by the number of years the site was observed. On average, burned sites (11 = 29) were surveyed 5.1 ? 2.2 (SD) years, with an occupancy rate of 0.88 ? 0.21. Unburned sites (11 = 24) were surveyed 5.3 ? 3.3 years with an average occupancy rate of 0.96 ? 0.08. Of the remaining sites, eight were surveyed numerous years as part of a study on
0
100
200
300 Kllorneters
Figure 1. Distribution of Mexican spotted owl survey areas in Arizona and N e w Mexico, 1997. Forty survey areas were located in Arizona (16 in the Coconino National Forest and 24 in the Coronado National Forest [Catalina, Pinaleno, Chiricahua, and Huachuca mountain ranges]), and 24 were in N e w Mexico (14 in the Gila National Forest and 10 in the Lincoln National Forest). All locations have an equal number of burned and unburned survey areas.
demography of Mexican spotted owls (Sean~anset al. 1999), but actual survey histories were not available. For the remaining three sites, the records did not show survey history but indicated only owl locations and dates. We conducted owl surveys within each site using three delineations of the site boundary. These were: (1) activity centers delineated by USFS biologists, based on historical use patterns and professional judgment; (2) a 400-m radius circle centered on the nest site or cluster of owl detections for each site; and (3) a I-km radius circle centered on the nest site or cluster of owl detections for each site. The 1-km radius circle (314.2 ha) was chosen because it was a convenient number that approximated the current guidelines for size and extent of Mexican spotted owl Protected Activity Centers (243 ha with irregular boundary; USDI Fish and Wildlife Service 1995). Because results differed little among the three types of boundary delineation and because the I-km circles had a more uniform size and shape than delineated activity centers and contained more grid points (see below) than the 400-111 circles, we will restrict our discussion here to results from the I-km circles and hereafter refer to them as "survey areas." Results of analyses involving the other two survey-area delineations are available in Jenness (2000).
Owl Surveys We conducted nocturnal calling surveys from 1 Mar. to 3 1 Aug. 1997, following established protocols (Forsman
1983, USDA Forest Service 1988). All surveys occurred between 30 min after sunset and 30 min before sunrise. Calling was discontinued in windy and stormy conditions. We called for owls from either fixed points or while walking along continuous transects, located so that calls were audible over the entire survey area. Each area was surveyed four times unless young spotted owls were observed outside the nest, confirming successful reproduction, before the fourth survey. If we heard a spotted owl at night, we returned to the area within 48 h to conduct a daytime follow-up survey to search for evidence of nesting. We searched for nests visually and, if unsuccessful, offered mice to the owls to attempt to locate the nest. In all cases, the same individuals surveyed both burned areas and their unburned counterparts. We assigned each survey area to one of four Owl Occupancy Ranks on completion of surveys. These were: ( I ) no owls detected; (2) one owl detected, but no conclusive evidence of pair occupancy; (3) a pair of owls detected, but no evidence of successful reproduction; and (4) young spotted owls observed outside the nest. This fourlevel ordinal categorization provides a reasonable basis for comparing occupancy or successful reproduction within established owl territories because each increasing level reflects increasing fitness. Quantitative measures of fecundity and survival over several years would better describe potential impacts on spotted owls (see Burnham et al. 1996, GutiCrrez et al. 1996, Raphael et al. 1996), but we did not have such data.
Determining Fire Severity, Cover Type, and Topographic Characteristics We sampled all burned survey areas for fire severity and dominant prefire vegetation type by systematically surveying a grid (186-m spacing, with an average of 91 2.9 points per survey area) randomly overlaid on each survey area map. Using each point as the center of a 10-m radius circle, we assigned the plot to one of four fire severity classes and one of four prefire vegetative cover types. We used criteria described by Wenger (1984) to distinguish between stand-replacing and stand-maintaining fires. Stand-replacing ("catastrophic") fires kill most vegetation within the fire boundary. Depending on the fire intensity and the vegetation density, stand-replacing fires may leave large amounts of dead vegetative material that increases the risk and intensity of future fires. Stand-maintaining fires generally burn close to the ground, killing the aboveground parts of grasses, shrubs, forbs, small trees, and burning accumulated ground fuels, without killing larger trees (Wenger 1984). We subdivided stand-maintaining fires into canopy-level and surface-level fires. Our fire severity classes were: (1) no evidence of fire within plot (unburned); (2) no evidence of canopy fire within the plot, but evidence of ground fire (ground fire); (3) evidence of patchy canopy fire within the plot (canopy fire); and (4) all canopy vegetation within the plot killed by fire (stand-replacement fire). We classified each site into one of four vegetation cover types (mixed-conifer, pine, pine/oak, other) based on the primary species composition of overstory and understory vegetation. Mixed-conifer sites were primarily Douglas-fir ( P s e ~ r d o t s u gmenziesii), ~~ spruce (Picetr spp.), or true fir (Abies spp.). Pine sites were characterized by forests of long-needled pines (Pit~usspp.). Pine/oak sites included mixed species stands of pine and oak (Quercris spp., primarily Q. gatnbelii). We also used 7.5' Digital Elevation Models (DEMs; Arizona State Land Department 1997, USDA Forest Service Geometronics Center 1997) to derive slope and aspect data and a crude index of topographic roughness within survey areas. We estimated topographic roughness by generating 20-nl contour lines over the survey area, clipping them to the survey area boundary, summing the total length of the clipped contour lines, and dividing by the survey area to yield a density of 20-m contour lines in meters per hectare.
+
gram BLOSSOM (Slauson et al. 1994) to examine the associations between occupancy ranking at burned sites and the severity and extent of fire, dominant cover type, and topographic characteristics in those sites. MRPP uses permutations of the actual data to calculate the probability that the observed grouping of observations could be due to chance, and thus can be used even when underlying distributions are non-normal. After a significant omnibus MRPP, we used univariate MRPP tests to identify specific variables associated with occupancy rank.
Results Description of Fire Extent in Owl Survey Areas On average, fires affected 55% of the area in burned sites (range 1-99%). This included 25% burned only by understory fire (range 1-73%) and 30% of the survey area burned by canopy-level fire (range 0-91%). On average, 14% of a survey area burned only partially into the canopy (range 0-44%), and 16% of the area burned at stand-replacing level (range 0-55%). Within the combined area of all our burned survey areas, fires burned 56% of the mixed-conifer type (24% at ground level, 17% in canopy, and 16% standreplacement), 66% of the pine type (26% at ground level, 18% in canopy, and 22% stand-replacement) and 54% of the pineloak type (34% at ground level, 8% in canopy, and 13% stand-replacement). Thirty-three percent of the burned portions were in mixed-conifer, 44% were in pine, and 14% were in pine/oak.
Associatioits Between Fire Presence and Owl Occupancy Rank Unburned survey areas tended to have more pairs and more reproductive pairs than did burned survey areas (Fig. 2). Coniparecl to the 33 burned sites, the 31 unburned sites had a higher proportion of occupied sites (84% versus 70%) and reproduction (16% versus 9%) in 1997. Of the 29 paired sites, 6 cases had a higher occupancy rank in the burned survey areas, 9 pairs resulted in ties, and 14 cases had a
Burned Sites 0 Unburned Sites
C
3
b
0.4
Statistical Analysis We analyzed our data in two steps. First, we compared occupancy rank between paired burned and unburned sites. Second, for burned sites only, we evaluated potential relationships between site characteristics and occupancy rank. We used a Test of Marginal Homogeneity (Mehta and Patel 2001) to compare occupancy rank among the 29 pairs of burned and unburned survey areas. We used a two-tailed test because fire could potentially degrade or improve owl habitat. We used the Multiple Response Permutation Procedure (MRPP [Mielke and Berry 19951) as implemented in pro-
fl,
Reproduction
Occupancy Ranks Figure 2. Relative proportions of burned and unburned Mexican spotted owl survey areas that had either no owls, a single bird, a pair, or confirmed reproduction. The numbers above the bars represent the actual numbers of survey areas at each response level, whereas the Y-axis represents the proportion of either burned or unburned survey areas at each response level.
higher occupancy rank in the unburned survey areas. These differences were not statistically strong ( P = 0.1 10). There was no marked difference in occupancy rank in relation to time since fire. Of the survey areas that burned in 1996 (one year before our surveys), six of eight were occupied, compared to 17 of 25 survey areas that burned 2-4 years before our surveys. Owls were present and reproducing at several sites that experienced severe fires. Fires burned >50% of each of the three burned survey areas with successful nests in 1997 (including stand-replacement-level percentages of 8, 3 1, and 32%), and the most severely burned survey area had a single owl in 1997. Thirty of the survey areas burned only once in the 4 years before our surveys, but three burned multiple times. At the latter sites, we found single owls at two sites and a nonreproducing pair at the other site.
Associations Between Fire Severity/Habitat Variables and Owl Occupancy Rank The omnibus MRPP test indicated that occupancy rank was significantly influenced by some variable or combination of variables ( P = 0.040). Subsequent univariate tests identified three variables as showing the most significant differences among the occupancy ranks: percent pine ( P = 0.003), percent mixed conifer ( P = 0.038), and percent of survey area unburned ( P = 0.075) (Table 1, Fig. 3). Owl sites where we detected no owls in 1997 contained more pine than sites where owls were present, and successful reproduction was associated with a higher proportion of mixed conifer (Fig. 3). The differences in percent of area burned relative to occupancy rank did not follow a biologically interpretable pattern (Fig. 3).
Discussion Our results were equivocal regarding the potential impact of fire on Mexican spotted owls. Although statistical significance was weak, our sample size was small, resulting in low power to detect a significant association between occupancy rank and fire. Further, we lack information on Table 1. For 33 burned sites, the probability that each fire severity and habitat variable did not differ among the four levels of occupancy rank (no owls, single owl, pair, reproduction) as determined by MRPP.
Variable Average slope in degrees % North-facing slope % East-facing slope % South-facing slope % West-facing slope % Unburned % Ground fire % Canopy fire % Stand-replacement fire O/n Pine % Pineloak % Mixed-conifer Topographic roughness
P-value 0.524 0.224 0.7 16 0.103 0.923 0.075 0.712
0.178 0.263 0.003 0.226 0.038 0.597
The o~nnibuslMRPP test using all predictor variables suggested that at least one variiible difrered among owl occupancy types ( P = 0.04).
Occupancy Rank ( n ) No Owls (10 )
@ Single (7) l
mPair (1 3) ~ R e D r o d u d o n(3'
% Pine P = 0.003
OIL
Mixed Conifer P = 0.038
Oh
-
Unburned P
= 0.075
Figure 3. Percent pine, mixed-conifer, and unburned predictor variables in 33 burned areas (314.2 ha each). Horizontal bars within boxes represent the median, the tops and bottoms of the boxes represent the 75th and 25th quantiles, and the whiskers represent the range excluding outliers and extremes. Outliers (values >1.5 box lengths from box) are displayed with the symbol "0" and extremes (values > 3 box lengths from the box) are displayed with the symbol "*".
spatial and temporal variability in owl occupancy rates, and our burned areas encompassed a wide range in both burn extent and severity. These factors further complicate interpretation of our data. However, point estimates of effect size (e.g., probability of occupancy 14% higher and probability of successful reproduction 7% higher in unburned sites) were suggestive of a biologically significant i~npact.Consequently, we caution against interpreting our results as indicative that fires d o not affect spotted owls. Our MRPP tests suggested that the percentage of pure pine in a survey area was most strongly associated with occupancy rank. Because fires burned greater proportions of the pine type in our survey areas, there may have been an interaction between fire effects and cover type. Previous evidence also is equivocal with respect to the effects of fire on Mexican spotted owls. Willey (1998) radiotracked three pairs of Mexican spotted owls before and after relatively low-intensity prescribed burns in Saguaro National Park, southern Arizona, finding that, at least in the short term, the owls tended to stay in the same gener.dl area after the fires. Bond et al. (2002) documented survival, site and mate fidelity, and reproductive success after large fires for 2 1 owls in 1 1 territories in California, Arizona, and New Mexico, finding high survival, site and mate fidelity, and reproductive success. Several Mexican spotted owls in the Gila National Forest continued to occupy telritories after prescribed natural fires that minimally altered stand structure (USDA Forest Service 1995), and there are numerous anecdotal observations of Mexican spotted owls occupying areas after wildfires and prescribed burns (Paul Boucher, Gila National Forest, 1997; Peter Stacey, Univ. New Mexico, July 8, 2001). Forest Science SO(6) 2004
769
Other anecdotal evidence demonstrates possible negative response to fire, such as Elliot's (1985) observation that at least one California spotted owl (S. o. occiderztcrlis) territory apparently was abandoned for several years after a highly destructive fire in 1977. Gaines et al. (1997) described impacts of 1994 wildfires on six northern spotted owl ( S . 0. CCILII-~IILI) activity centers in eastern Washington, noting a small decrease in the number of reproductive pairs on these sites, and an increase in the number of unoccupied sites the year after the fires. Bevis et al. (1997) discussed two pairs of radio-tagged northern spotted owls in south-central Washington that stayed near their territories after a 1994 wildfire but shifted their primary activity to lighlly burned 01. unburned areas. One female owl in that study was found dead in an emaciated condition 2.5 months after the fire, leading to speculation that the fire may have reduced prey availability, and her mate disappeared over the winter. Two new owls occupied the territory one year after the fire, suggesting that, by then, the te~ritorycould support owls despite the fire. The potential for widespread stand-replacing fire, identified as a leading factor endangering the Mexican spotted ow1 by USDI Fish and Wildlife Service (1993, 1995), has not abated during the last decade. For example, Sheppard and Farnsworth (1997) estimated that at least 10 Mexican spotted owl territories were affected by fire in A r i ~ o n aand New Mexico during the 1994 fire season alone. They further estimated that >20,234 ha of owl habitat experienced standreplacing wildfire from 1989-1994. Severe wildfires have occurred in virtually every year since 1994, with peaks in 1996 and 2002. Thus, stand-replacing wildfire remains a significanl- threat to owls and their habitat in both the short and long-term. We recognize that most wildfires burn in a patchy nature and leave pockets of useable habitat for owls. We further recognize that some owls locate and use these patches ancl thus are relatively resilient to wildfire, at least in the short term. Nevertheless, we continue to view the potential cumulative loss of habitat to wildfire as a major threat to the owl, especially given lack of data on long-term responses of owls to fire. Our concern over loss of owl habitat to wildfire is exacerbated by several factors. One is the current structure of southwestern forcsts. Although owl habitat as we know i t today is inherently lire prone, with high tree densities, canopy cover, and fuel loadings (Ganey and Dick 1995), we suspect that those conditions were not continuous on the landscape under natural fire regimes (Marshall 1963, Co\lington et al. 1994). Such conditions now are relatively continuous on the landscape, raising the specter of more, larger, and more intense wilclfires. This creates greater potential for widespread loss of owl habitat. Further. much of the currently occupied ow1 habitat occurs in canyons feati~rjngcool rnicrosites and rock outcrops. These areas may have functioned as fire refugia (.~ensu Camp et al. 1997) in historic times, escaping some of the frequent fires and allowing longer fire-free periods for stand development. With increasing intensity of wildfires (Moody et al. 1992) and greater continuity of heavy fuels.
the potential for fire to spread into these refugia would appear to be higher as well. This problem is further exacerbated by the fact that the southwest is currently experiencing a severe drought of unknown duration, further increasing the potential for large and intense wildfires (Swetnam 1990, Swetnam ancl Betancourt 1990). Finally, the seasonality of fire in the southwest has changed. Most naturally occurring fires in Arizona and New Mexico occur during the monsoon season between July and September as a result of lightning strikes during thunderstorms (Sackett et al. 1994, FulC et al. 1997, Brown et al. 2001). This is a rainy season in the southwest when f ~ ~ e l moistures are often high and precipitation is likely to fall concurrent with or soon after the lightning strikes, both factors which can moderate fire behavior and extent. In contrast, many large wildfires in recent years have been human-caused, and have occurred in May and June (historical data on timing, size, and cause of large fires, wwu;.fs,fed.us/r3/firelswainfols\vainfo.htm,Sept., 2003). This is a dry period in the southwest, Palling between periods of winter precipitation and summer thunderstorms. This season is characterized by low fuel moistures, high winds, and little chance of precipitation to moderate fire behavior. These conditions, coupled with high and relatively continuous fuel loadings as a result of fire suppression, favor extreme fire behavior and large fires. We also suspect that fire may affect owls differently based on whether i t burns in corelnest areas where i t may cause changes in the microclimate and cover necessary for raising young, or foraging areas where it may affect prey availability and diversity. We were onable to analyze this possibility because of a lack of sufficient historical nesting locations in our dataset, but we feel it could have important management implications and should be considered in fulure research or management actions.
Management Implications The recovery plan for the Mexican spotted owl (USDI Fish and Wildlife Service 1995) recommended proactive fuels management to reduce continuity of fuels on the landscape. We support that recommendation. The plan further recommended that initial fuels-reduction efforts focus on areas not currently occupjed by owls (Protected Activity Centers or PACs) because of uncertainty over the effects of lhose treatments on owls. We also believe this recommendation is appropriate. We agree with the plan's conclusion that treatment of areas outside of PACs can provide considerable protection to those PACs by allowing managers lo reduce continuity of' fuels on the landscape, thus reducing the potential for fires to start elsewhere and spread into PACs. Because owl PACs comprise only approximately 2.1 % of National Forest lands in Arizona and New Mexico (USDA Forest Service, unpublished data), we further believe that this approach leaves open to management more area than land management agencies can reasonably treat in the near Suture, and thus should not prevent managers from addressing fuels-management issues in most areas. We recognize, however, that unique situations may exist requiring
treatments LO proceed in PACs. One exaniple is cases where owls occur in the wildland-urban inlesface, and treatment of those areas is necessary to reduce fire risk to rural and urban areas of human settlement. Other examples undoubledly exist. We recommend that managess evaluate these sittiations on a case-by-case basis. Ideally, where fuels reduction treatments proceed in PACs. associated effecls on Mexican spotted owls and key ecological linkages (e.g.. prey and imporlant habitat variables) should be determined by incorporating those treatments in a management experiment using a credible scientific design, e x a n ~ i n i n glongterm survival and fecundity, and including a rigorous multiyear monitoring program. However, such an effort may prove prohibitively expensive. Thesefore, as an alternalive, we suggest that sesponse of owls be monitored using systematic and standardized research designs that will allow results from vasious areas to be included in a meta-analysis evaluating effects of fuels treatments on owls, their habitat, and important prey species.
Literature Cited AI
